Daytime Sleepiness linked with Sleep and Mental Disorders

Daytime sleepiness is one of the principal symptoms of a sleep disorder. Prevalence of daytime sleepiness has been reported to range from 0.5% to about 40% (1-8). Inconsistent definitions (e.g., getting too much sleep, falling asleep in the day), study populations (males, middle aged or elderly population) and reported frequency of symptoms, have contributed to this variable prevalence rate.

The etiology of daytime sleepiness involves various sleep disorders such as obstructive sleep apnea syndrome (9-11), heavy snoring (12, 13) and narcolepsy (14).However, the presence of daytime sleepiness can be explained by other disorders, such as an insufficient sleep syndrome (15) and/or poor sleep hygiene.

Sleepiness is involved in approximately 16% of motor vehicle accidents in England (16).

Moreover, it has been suggested that half of the work-related accidents and a quarter of home-based accidents are caused by sleepiness (17).



Telephone interviews were used to ascertain and describe sleep habits, sleep-related symptoms and psychiatric disorders of the British population. To this purpose, a representative sample was selected from the non-institutionalized population aged 15 years and over using a stratified probabilistic approach.

The sample selection consisted of a twofold process:

  • first, with respect to the geographic distribution of the UK population as described by the 1991 census survey and
  • second, with respect to age and gender of each household member 15 years and over, using the Kish selection method (18) which specified the individual to be interviewed.

Non English speakers, individuals with an illness that prevented the feasibility of an interview or those who suffered from a hearing or speech impairment were excluded from the survey.

Persons who refused to participate after two attempts or who gave up before having completed half the interview were classified as a “refusal.” Selected phone numbers with no answer were dialed again at least ten times at different hours and different days, which included both weekdays and weekends, before being replaced. A total of 8686 phone numbers were attempted, of which:

  • 57.2% resulted in completed interviews,
  • 13.5% in refusals,
  • 12.6% were not in service or were business lines,
  • 9.4% went unanswered after ten attempts, and
  • 7.2% met an exclusion criterion or could not be reached during the survey.

Of the 6249 subjects eligible to participate in the survey, 4972 agreed to be interviewed (79.6%). The ages ranged from 15 to 100 and included 2894 women and 2078 men.


Telephone interviews were performed by a company specializing in nationwide telephone surveys (BPS Teleperformance, Birmingham) from June to July 1994 and from September to October 1994.

Forty interviewers participated in the study. They were inexperienced in psychiatric diagnosis and sleep disorders but were specially trained to use the Sleep-Eval system. The average training time was about two half days. Two supervisors monitored the interviewer team daily to ensure that the interviews were correctly conducted and data entered properly. A large part of their duty consisted of listening to the interviews as a third party, while the interviewers were unaware of the moment the supervisor would listen to an interview in process.


The expert system, Sleep-Eval, is a non-monotonic, level-2 expert causal reasoning system (19), designed to provide homogeneous and standardized evaluations.

Sleep-Eval has been validated (20, 21, 22) and was found to be an appropriate tool by which to manage both the epidemiological study and the administration of the questionnaire. The system was tested within several designs:

  • In face-to-face settings, the kappa ranged from 0.45 for psychotic disorders (23) to 0.98 for depressive disorders (20).
  • The diagnoses made by the Sleep-Eval system, used by lay interviewers, was also tested against the diagnoses of two clinician psychologists using telephone interviews with subjects of the general population. Good test-retest results and agreement between the interviews was found (unpublished data).
  • In addition, the reliability of the system’s reasoning was verified by reconstructing each step of the decision-making process as part of the data analysis process: The algorithm to achieve a DSM-IV (24) or an ICSD-90 (25) diagnosis has been recreated from the responses provided to questions. Then, results from this algorithm were compared to the diagnoses given by the expert system.

The “Sleep-Eval” system is composed of three modules.

  1. The first reconstructs the subject selection procedure described by Kish (18).
  2. The second executes the logical reasoning required for customized administration of the questionnaire. The system first loads a set of standard questions for the entire population. Based upon the data collected from the responses, the system explores a series of diagnostic hypotheses (causal reasoning process) to be confirmed or rejected through further questioning or deductions (non-monotonic, level-2 feature). For example, a subject who had responded negatively to hypersomnolence screening questions but stated feeling “a lot” sleepy during the day would also activate questions to explore a sleep or breathing related disorder. The system contains all the questions required for the entire diagnostic description. The interview is complete when the system have exhausted all the diagnostic possibilities. The general structure of the questionnaire is similar to that used in traditional tools, beginning with general questions on sociodemographics and sleep habits and leaving emotionally loaded questions until the later part. Each interview is adapted depending on the answers provided by the subject on a series of key questions. Positive or negative answers on these questions fire a reasoning process similar to the differential diagnosis applied by a psychiatrist. The power of such a tool resides in its ability to use all the information provided by the subject in order to achieve or eliminate a diagnosis. The differential process is based on a series of key rules allowing or prohibiting the co-occurance of two diagnoses as defined in each classification (DSM-IV and ICSD-90). The strength of this tool rests in its ability to provide reliable, objective assessments. Namely, the presence of a symptom is based upon the interviewee’s responses rather than on the interviewer’s judgment. This approach has been proven to yield a better agreement between lay interviewers and psychiatrists on minor psychiatric disorder diagnoses (26).
  3. The third module manages and classifies the files according to interview outcome (i.e., subjects with or without sleep or psychiatric disorders) or whether a number resulted in a refusal to participate, was dialed up to ten times unsuccessfully, or remained open and had to be dialed again.

Once a file is classified, it is closed and the interviewers have no longer access to it.

Questions are read out by the interviewer as they appeared on the screen. These questions were either closed-ended (e.g., yes/no, 5-point scale, multiple choice) or open-ended (e.g., duration of symptom, description of illness).

The questionnaire included in the “Sleep-Eval” system covers several topics. In this study, responses to questions related to daytime sleepiness were analyzed in relation to the following major categories: sociodemographic information; sleep/wake schedule; quality of nocturnal sleep or longest sleep period; symptoms or behaviors occurring in sleep; current medication intake; daily intake of tobacco, caffeine, alcohol, and other drugs and their quantity; medical consultations, hospitalizations and treatments during the past 12 months; DSM-IV mental disorders and ICSD-90 sleep disorders.


Sleepiness was assessed through the question “Do you feel sleepy during the day?” Subjects responded on a five point scale:

  • Not at all
  • slightly
  • moderately
  • a lot
  • greatly

If appropriate, subjects specified:

  • the period of the day where sleepiness was the most severe and
  • how long it lasted.

Study groups were defined as follows for the purposes of analyses:

  1. Severe daytime sleepiness (SDS): subjects who reported being “a lot” or “greatly” sleepy on a daily basis for a period of time lasting at least one month.
  2. Moderate daytime sleepiness (MDS): subjects who reported being “moderately” sleepy on a daily basis for a period of time lasting at least one month.
  3. No daytime sleepiness (NDS): subjects who reported being “not at all ” or “slightly” sleepy.

Subjects who reported feeling “a lot”, “greatly” or “moderately” sleepy on a daily basis for a period lasting less than one month were excluded from the analyses. This was the case for 16 subjects.


A weighting procedure was applied to compensate for disparities between the sample and the standard population with respect to the age and gender distribution (1991 UK census of the non-institutionalized population aged 15 years and older).

Results are presented with weighted percentages and 95% confidence intervals (95% C.I.) when appropriate. Bivariate analyses were performed using Chi square statistics and the Bonferroni correction for p values.

ANOVA with post hoc multiple comparison test (Duncan’s multiple range test) were also used to compare mean sleep duration and sleep efficiency.

Predictive factors of daytime sleepiness were identified with the help of logistic regression procedures using a method of quasi-sequential selection (forward) with the likelihood-ratio (LR) test to identify variables to be removed from the model (27). Colinearity between variables (i.e., information redundancy) was checked prior to make the logistic regression. The cut-off point of entry (PIN) was fixed at .05 and the cut-off point of exclusion (POUT) at .10. Reported differences were significant at the .05 level or less.

Analyses were performed with the SPSS computerized statistical package (SPSS 6.1).


The final sample consisted of 4956 subjects, with an age ranging from 15 to 100 years. The proportion of women in this sample was 52.3%. Severe daytime sleepiness (SDS) was reported in 5.5% (95% CI 4.9% to 6.1%) of the sample, and moderate daytime sleepiness (MDS) in 15.2% (14.2% to 16.2%) of subjects. More women [6.6% (5.6% to 7.6%)] than men [4.4% (3.6% to 5.2%)] reported SDS (P < .001). The highest SDS prevalence was found in middle aged subjects (35 to 44 years old) and highest prevalence of MDS was observed in elderly (table 1). Most SDS and MDS subjects had experienced daytime sleepiness for more than 24 months prior to the interview (41% and 32%, respectively). Approximately one fifth of SDS and MDS subjects (22%) experienced daytime sleepiness for one to six months. For both SDS and MDS subjects, sleepiness was more important in the afternoon than at any other time during the day (SDS= 76.1%, MDS= 62.5%, P < .001). Occupational or social repercussions related to daytime sleepiness were reported by 41.2% of SDS and 14.8% of MDS (P < .001).


Subjects who reported sleeping 6 hours or less per day or an extended sleep latency (< 30 minutes) were more likely to be SDS or MDS (table 2). Mean sleep duration was 6.08±1.54 hours in SDS subjects compared to 6.57±1.37 hours in MDS and 6.82±1.21 hours in NDS subjects (P < .001). Furthermore, sleep efficiency (sleep duration / time spent in bed) was significantly lower in the SDS group (mean: 83.61%±15.33%) compared to MDS (mean: 87.16%±12.58%) and NDS (mean: 89.61%±10.71%) groups (P<.001). Reporting dissatisfaction with irregular bedtime and/or morning awakening hours was associated with both a higher report of SDS and MDS (table 2). Additional hours slept on weekends or days off did not produce significant difference in daytime sleepiness. Finally, napping more than once a week had an odds ratio five times higher for SDS (OR=5.5 [4.3-7.1]) and an odds ratio four times higher for MDS (OR=3.6 [3.0-4.3]) than NDS (P<.001). Longer napping (< 1 hour) appears more prevalent in SDS subjects (28.4%) than in MDS subjects (9.3%; P < .001).


Daytime alcohol consumption was not associated with daytime sleepiness. However, alcohol intake in bed prior to sleep was associated with a two times higher risk of being SDS (OR=2.1 [1.4-3.0]) than NDS (P < .001). The quantity of daily cigarettes did not differ between groups, however the rate of SDS was higher in daily smokers (7.7%) compared to nonsmokers (4.8%; P < .001). Only when subjects reported drinking at least seven cups of tea or coffee per day, was the rate of SDS greater than NDS (10.6% vs 5.9%; P < .005). Not surprisingly, subjects using sleep-enhancing medications were more likely to have SDS (OR=3.7 [2.4-5.5]) or MDS (OR=1.6 [1.1-2.3]) than NDS (P < .001). Subjects taking anxiety-reducing (OR=6.7 [3.7-12.1]; P < .05) or antidepressive medications (OR=4.5 [2.5-8.3]; P < .05) were also more likely to have SDS.


Daytime sleepiness was observed in subjects reporting various sleep disorders. Both SDS and MDS subjects were more likely than NDS to be diagnosed with a Mood Disorder associated with sleep disturbance, Psychophysiological Insomnia, Obstructive Sleep Apnea Syndrome or Insufficient Sleep Syndrome. Restless Legs Syndrome was more often observed in SDS subjects than in NDS. Narcolepsy, using minimal criteria described in ICSD-90, was identified in 0.04% of the sample. In addition, some parasomnia diagnoses were more often observed in SDS subjects, including: Confusional Arousals, Nocturnal Leg Cramps, Nightmares and Sleep Paralysis (table 3).


Subjects with an SDS more often had a DSM-IV Depressive Disorder (17.4%; P < .001) or a DSM-IV Anxiety Disorder (19.4%; P<.001) compared to MDS subjects (8.7% and 9.6% respectively) and NDS subjects (3.3% and 4.2% respectively). DSM-IV Bipolar Disorders were more often found in SDS (6%) and MDS (3.9%) than in NDS subjects (1.5%, P < .001). Similarly, histories of a previous mood or anxiety disorder were reported more often by SDS subjects (12% and 12.6% respectively) compared to MDS (6.7% and 6.8% respectively) and NDS subjects (4% and 3.7% respectively;P < .001).


SDS and MDS subjects had consulted a physician during the last 12 months (76.3% and 67.9% respectively, P < .001), were hospitalized in the last year and/or treated for a physical disease (20.2% and 15.6% respectively, P < .001) more frequently than NDS subjects (58.7% and 9.5%). More subjects in the SDS group consulted a physician six or more times (38.1%) than in the MDS (24.5%) and NDS groups (17.0%, P < .001). SDS subjects reported more often to be treated for a physical disease (27.4%) than NDS subjects (12.8%, P < .001). The most frequently treated diseases in SDS subjects were: arthritis (17.8% vs 5.1% in NDS), heart diseases (5.1% vs 2.2% in NDS) and obstructive airway diseases (3.1% vs 1.5% in NDS).


Approximately 60% (n=3014) of the sample regularly operated machinery or a motor vehicle. This proportion did not significantly differ between the groups classified on the basis of daytime sleepiness, yet twice as many SDS subjects reported having a road or machine accident (9.3%, P=.02), and/or having received one or more tickets in the previous 12 months (6.9%, P=.02) compared with the NDS group (4.7% and 3.4% respectively). The problems associated with daytime sleepiness included a tendency to sleep while driving which extended to the need to be accompanied to prevent falling asleep at the wheel. This occurred significantly more often in SDS than NDS subjects (12.2% versus 4.4%; P < .001).


Thirty-one variables were introduced into the models used to predict daytime sleepiness: age, gender, presence of children, quality of sleeping environment, presence of snoring, breathing pauses, sleep agitation, excessive limb movements in sleep, drinking an alcoholic beverage before going to sleep, regularity of sleep-wake schedule, falling asleep while reading or watching television, dissatisfaction with the wake up time or time of falling asleep, leg pain accompanied with muscular stiffness or hardening during sleep, level of sleep satisfaction, depth of sleep, sleep latency, sleep duration, smoking, daily intake of tea or coffee, consumption of psychotropic medication, body mass index, presence of an upper airway disease, heart disease, arthritis disease, naps, difficulties in initiating or maintaining sleep, early morning awakenings with inability to resume sleep, non restorative sleep, Depressive or Anxiety Disorders. SDS was positively associated with being a woman, being 25 to 44 years old, daytime napping at least twice weekly, reporting a global dissatisfaction with sleep, a complaint of nonrestorative sleep or difficulties in maintaining sleep, a diagnosis of Depressive Disorder, a report of breathing pauses in sleep, a daily intake of seven cups or more of tea or coffee, a sleep latency longer than 30 minutes, a dissatisfaction with the wake up time, leg pain accompanied with muscular stiffness or hardening during sleep, and falling asleep while reading or watching television (table 4). MDS was positively associated with being a woman, daytime napping at least twice weekly, a complaint of a nonrestorative sleep or difficulties in maintaining sleep, having an arthritic disease or a heart disease, falling asleep while reading or watching television, a dissatisfaction with wake up time or bedtime hour, an irregular wake up time, with an excessive limb movements during sleep, and a global dissatisfaction with sleep (table 4).



Our unusual methodology might raise some interrogations. Successful attempts at using computers for epidemiological surveys have been reported (28). Interviews performed over the telephone have been found to provide a reliability of psychiatric diagnoses comparable to that of face-to-face (29-32) with the additional advantages of lowering study costs and cutting the time needed to complete the study. Telephone interviews are thought to diminish the discomfort involved in responding to emotionally charged questions by the partial anonymity offered by the telephone and, consequently, may reduce bias and increase the validity of responses (32). Compared with traditional paper-pencil questionnaires, the use of an expert system shortened interviews for subjects without mental health problems by eliminating irrelevant questions; provided practically no missing data; suppressed manual data entry and the risk of error due to data transcription or encoding; ensured the standardization of the administration.

To verify the reliability of this instrument, the reasoning process of the Expert system was verified again at the time of data analyses by reconstructing its algorithms and verified according to the DSM-IV and ICSD-90 classifications. In our opinion, the reasoning process of all Expert Systems used in Medicine should be verified at each step of the decision-making process even if the system has 100% diagnostic agreement with clinicians. After this, the only possible remaining bias would be the inability of the system to recognize some infrequent pathologies.

Another limit could rest in the portion of households without telephone (7%) in the UK. This rate differs somewhat with other industrialized countries (e.g., 6% in France and 2% in Canada). However, this is low enough to not obscure valid inference (33, 34).

The absence of laboratory assessment deserves some discussion. In clinical studies, the Stanford Sleepiness Scale (SSS) (35) was one of the first scales allowing a wide-ranging assessment of daytime sleepiness. Although the SSS is still used, the Multiple Sleep Latency Test (MSLT), a laboratory sleep behavior measure, is considered to be the “gold standard” for assessments of daytime sleepiness. Some studies have also used pupillometry (36) as a biological measure of daytime sleepiness. Unfortunately, these measures cannot be applied in epidemiological studies unless some subjects are further asked to participate in laboratory evaluations. Self-reported daytime sleepiness scales, such as the Epworth Sleepiness Scale (ESS) (37) and the Sleep-Wake Activity Inventory (38) were recently developed. These types of scales, like the MSLT, reflect the sleep propensity but not the subjective feeling of being sleepy (39). This distinction is of interest as it may explain why some patients rated themselves as being sleepy while objective measures were not consistent with these reports (40, 41) and conversely, why some did not report daytime sleepiness but displayed several symptoms of sleep propensity in the day. Many authors have hypothesized that subjective and objective measures of daytime sleepiness evaluate different aspects of the sleepiness (42, 43) and that sleepiness is not unidimensional (44).

Finally, the diagnostic findings are based on self-reported symptoms without the recourse of polysomnography and sleep disorder diagnoses were established on the basis of minimal criteria as described in the ICSD-90. This may pose a problem for some diagnoses such as breathing related sleep disorders. However, the rate of Obstructive Sleep Apnea Syndrome found in this sample was 1.5% in women (95% C.I. 0.8% to 2.2%) and 3.5% in men (95% C.I. 2.4% to 4.6%) which is similar to that reported by Young et al. (45) where polysomnography was used (2% in women and 4% in men within the same age range).


The evolution of knowledge from epidemiological surveys on daytime sleepiness has been stagnant due to the use of inconsistent diagnostic criteria. Early epidemiological studies (4, 6) assessed daytime sleepiness as a dichotomous variable (presence or absence of the symptom).

Later, others used frequency of the symptom (1, 8) however, duration of the symptom was usually omitted. The Ford & Kamerow’s study (2) set the duration at a period of two weeks or longer. The terminology employed also varied between studies: some asked about “falling asleep during the day” or “sleeping too much”, others about “excessive daytime sleepiness” or “hypersomnia”. Daytime sleepiness was usually assessed using only one or two questions which prevented the possibility of making comparisons with other daytime sleepiness manifestations within the same study. Moreover, in different studies, different aspects of daytime sleepiness were queried, making comparisons between epidemiological studies difficult.

In this survey, daytime sleepiness had lasted one month or longer, and was measured using a subjective assessment based upon its severity. In addition, some items allowed assessment of the individual’s sleep propensity, i.e., tendency to fall asleep when intending to stay awake. Subjective daytime sleepiness was a common symptom in our sample, affecting 5.5% in its severe form and 15% in its moderate form. Hypersomnia, according to the DSM-IV and ICSD-90, is a sleep disorder characterized by excessive sleepiness as evidenced by an extented sleep duration or many daytime naps. This last disorder is rare: In our sample, only 1.7% of subjects slept 10 hours or more and less than 1% met the complete description of hypersomnia. Gender effects have been reported in some studies (8) but not others (1, 2, 6). Similarly, younger cohorts were found at higher risk for daytime sleepiness in some studies (2, 7, 8) whereas elderly were found at higher risk in others (1). In our study, severe daytime sleepiness occurred slightly more often in women than men, and middle aged subjects were twice as likely to suffer from SDS than all other age groups. Moderate daytime sleepiness mostly involved elderly.

Contrasted with previous epidemiological surveys, our study considered several aspects related to daytime sleepiness, including sleep-related and mental disorders, plus several factors which were believed to contribute to the reporting daytime sleepiness. Our results showed that sleep dissatisfaction and insomnia complaints considerably increased the risk of reporting both moderate and severe daytime sleepiness, with the risk for moderate daytime sleepiness lower than for severe daytime sleepiness.

Other factors related to sleep propensity, such as napping or falling asleep while reading or watching television, also increased the likelihood of reporting both moderate or severe daytime sleepiness. In our sample, napping at least twice weekly was reported by 45% of the severe daytime sleepiness group and by 31.8% of the moderate daytime sleepiness group. Falling asleep while reading or watching television was reported in similar proportion by both groups (about 38%). Indicators of sleep propensity were found in 65% of subjects with severe daytime sleepiness and 56% of subjects with moderate daytime sleepiness. A sizable proportion of subjects saying they were not sleepy during the day also had some indicators of sleep propensity (32%).

Some somatic disorders (such as hypertension, arthritic diseases, diabetes) (5, 7) or poor health status (46) were found to be related to daytime sleepiness. In our study, arthritis and heart disease were predictive of moderate sleepiness but not of severe daytime sleepiness. These two diseases are commonly found to be related to poor sleep quality and nocturnal complaints and may produce daytime sleepiness (47).

Not surprisingly, subjects with a depressive disorder were at a higher risk of reporting severe daytime sleepiness as the presence of hypersomnia nearly every day is one of the associated criteria for either Major Depressive Disorder or Dysthymia Disorder. The association between daytime sleepiness or hypersomnolence and mood disorders has been outlined many times in clinical research (48, 49). Also, it has been demonstrated that anxiety disorders involve disrupted sleep and daytime sleepiness (50).

Our results showed that daytime sleepiness was frequently related to the report of nocturnal symptoms, most notably, difficulties maintaining sleep (nocturnal awakenings) and non restorative sleep (i.e., feeling that the sleep is unrefreshing even if the sleep duration was normal). Disrupted sleep has been previously reported to be a more significant cause of daytime sleepiness than other nocturnal symptoms (51). Daytime sleepiness was found associated with various sleep disorders which may include this symptom. The most frequent were Mood Disorder (Depressive or Bipolar Disorder) associated with sleep disturbance, Obstructive Sleep Apnea Syndrome, Restless Legs Syndrome and Insufficient Sleep Syndrome which are commonly believed to produce daytime sleepiness (52, 53).

Previous studies have shown that daytime sleepiness is one of the cardinal symptoms of obstructive sleep apnea (10, 11). Other studies have reported that nonapneic heavy snorers are more likely to complain of daytime sleepiness (13, 54). Furthermore, it has been demonstrated that an abnormal increase in respiratory efforts during sleep may bring several brief arousals, usually ignored when examining EEG with long epoch scoring, for a subset of idiopathic hypersomnia subjects (55, 56). In our study, subjects with breathing pauses during sleep had twice the risk of complaining of severe daytime sleepiness while this risk was not identified in subjects with moderate sleepiness. In our sample, obstructive sleep apnea syndrome was found in 6.4% of subjects with severe daytime sleepiness and 4.1% of those with moderate sleepiness. Obstructive airway diseases were reported by 3% of both groups. Breathing disorders explained only a small proportion of daytime sleepiness observed in the general population. However, an individual with symptoms affecting normal breathing in sleep was more likely to report some degree of daytime sleepiness (9, 45, 57).


Daytime sleepiness may reflect various causes:

  • It can be the primary symptom such as in narcolepsy;
  • it may be induced by lifestyle factors such as work conditions, poor sleep hygiene habits, or psychotropic consumption;
  • it can be secondary to various psychiatric, breathing disorders; or
  • it can be the result of circadian rhythms perturbations.

Thus, daytime sleepiness is most often caused by specific factors which can be easily identified by the physician and consequently treated.

Consequences of daytime sleepiness may affect several areas of functioning: at work, in social or marital life, and provoke decreased concentration or memory problems (51, 58).

In addition, a sleepy individual is at greater risk to have road, work-related or home-based accidents (17, 59).

It is of importance to underline that we found twice as many subjects with severe daytime sleepiness having a road or machine accident in the previous twelve months as those individuals with no daytime sleepiness.

The high prevalence of daytime sleepiness clearly indicates that this is a very important public health problem requiring preventive and educational initiatives on a large scale.

The socioeconomic impacts of daytime sleepiness should appeal to public health authorities.


Content of this page is extracted from:
Ohayon MM, Caulet M, Philip P, Guilleminault C, Priest RG. How sleep and mental disorders are related to complaints of daytime sleepiness. Arch Intern Med 1997;157:2645-52.